ai-systems

Files (1)

• SKILL.md

SKILL.md (first lines)

# AI Systems — Founder OS

Builds production-grade AI workflows and improves output quality. Two layers: the Meta-Thinking layer (getting better outputs from any prompt) and the Production Architecture layer (building reliable AI systems).

---

## Layer 1: Meta-Thinking Prompts (Output Quality)

### The 3 Stuck Points

| Stuck Point | Symptom | Root Cause |
|-------------|---------|-----------|
| **Blank Chat Paralysis** | Know the outcome, can't frame the request | Haven't given AI enough context to work from |
| **Generic Output Loop** | Clear question, vague/textbook response | Right question, wrong altitude or missing constraints |
| **Solving the Wrong Problem** | Polished output, but wrong answer | AI optimized for what you said, not what you need |

### The 9 Meta-Thinking Prompts

**For Stuck Point 1 — Let AI interrogate you first:**

**Prompt 1: Context Extraction**
```
Before answering, ask me 7-10 questions that would help you understand
my situation well enough to give strategic, not generic, advice.
Don't answer yet — just ask the questions.
```

**Prompt 2: First Principles Reset**
```
Break this down using first principles.
What am I assuming that should be questioned?
What's the real problem underneath the problem I described?
```

**Prompt 3: Gap Identifier**
```
What information do you need from me to move from a generic framework
to something implementable for my specific situation?
List exactly what's missing.
```

---

**For Stuck Point 2 — Fix the altitude:**

**Prompt 4: Altitude Diagnostic**
```
Am I asking for strategy when I need tactics?
Or tactics when I need strategy?
What altitude should this conversation be at, and why?

Altitude levels:
- 30,000 ft: Vision ("What should we become?")
- 20,000 ft: Strategy ("How will we win?")
- 10,000 ft: Tactics ("What specific actions?")
- Ground: Execution ("Step-by-step instructions")
```

**Prompt 5: Missing Context Probe**
```
What would make your last response completely wrong for my situation?
What assumptions did you make that I should validate?
```

**Prompt 6: Inversion Technique**
```
What would make this the worst possible plan?
List every way this could fail, then help me address each failure mode.
```

---

**For Stuck Point 3 — Stress-test the output:**

**Prompt 7: Usefulness Audit**
```
Rate what you just gave me 1-10 for actual usefulness — not theoretical quality,
but practical usefulness for a founder at [my stage] trying to [specific goal].
What specific changes would make it a 10?
```

**Prompt 8: Devil's Advocate**
```
Play devil's advocate on your own response.
What are the 3-5 strongest criticisms of this approach?
Don't defend it — genuinely challenge it.
```

**Prompt 9: Rejection Test**
```
Imagine I'm the CEO reviewing this output. I'm about to reject it.
What are the most likely reasons I'd reject it?
Then rewrite it to survive that rejection.
```

### Power Stack: Run in Sequence

For any high-stakes output (investor pitch, pricing strategy, positioning):
1. Start with **Prompt 1** (Context Extraction) — let AI ask you questions first
2. After answering, run **Prompt 4** (Altitude Diagnostic) — make sure you're at the right level
3. After getting output, run **Prompt 9** (Rejection Test) — harden the output

This sequence turns a 20-minute prompt session into a 45-minute session that produces work worth 10x more.

---

## Layer 2: Production Architecture (Building AI Systems)

### The 7-Step Production Framework

For building reliable Claude Projects, CustomGPTs, or any AI workflow that runs repeatedly.

**Step 1: Brainstorm-to-Blueprint (4 phases)**
```
Phase A — Problem Mapping:
What specific task does this system need to do?
What's the input? What's the expected output?
What does success look like (measurable)?
What does failure look like?

From The13thNode/VibeCorp_PromptCEO, MIT.

More from The13thNode/VibeCorp_PromptCEO